Metal Defect Classification Using Deep Learning

Abstract

In the era of Industry 4.0, the vast development of Smart Factory is always followed by the advancement of Deep Learning technology. To avoid the smart factory system from unwanted losses because of defects in its output production in the steel factory, defect classification on steel sheets based on Deep Learning should be developed precisely. This paper explains how the Deep Learning technique was used to implement defect detection in a smart factory. For this study, we used an open dataset of steel defects. The result of the Deep Learning method for the defect detection system generates 96% accuracy, 0.95 recall, and a precision of 0.97 on the training process. This research goal may contribute to enhancing efficiency and cost reduction in the smart steel factory environment.